A novel Kaniadakis entropy-based multilevel thresholding using energy curve and Black Widow optimization algorithm with Gaussian mutation

In this research, Kaniadakis entropy (KE) derived from the energy curve is adopted to construct an objective function for the thresholding of images at various levels. In addition to the histogram's property, the energy curve maintains the image's spatial contextual information. This additional data aids in the threshold selection process, resulting in a more accurate segmented image. To optimize the objective function, a new Black widow optimization with a gaussian mutation algorithm (BWOG) is also proposed in this paper with enhanced population diversity by incorporating an additional stage of powerful Gaussian mutation and random allocation of solutions using a levy flight mechanism in BWO. The proposed Kaniadakis entropy-based multilevel thresholding selection using energy curve and Black Widow optimization algorithm with Gaussian mutation (BWOG-KE) is performed on both grayscale and color images of different modalities and dimensions. Based on the quantitative measures: PSNR, the BWOG-KE is found superior to existing well-known methods. The results proposed method are further compared with minimum cross-entropy (MCE) based and Kapur's entropy-based thresholding and found a significant level of dominance over them.